Cognitive Neuroscience Lecture 17: Memory 1
Learning and Memory * Two kinds of knowledge: ** Hard-wired ** Learned (memories) * Learning is the process of acquiring new information * Memory is the persistence of learning in a state that can be revealed at a later time * Memory systems reflect learning systems 'Fundamental task of a memory system: preserve information over time Why would you want to preserve information? That way, you can use past to predict future using the internal model created with your memory. ''' * Increase our ability to function in present and future based on past experiences * Support development of an internal model of the world that improves prediction What does a system have to do to support these functions? * Encode, store, retrieve * Allow stored information to be accessed flexibly to adapt to a changing environment Distinctions in memory (lecture outline) * Time scale of storage: ** Sensory ** Short term (working) ** Long term *** Declarative (explicit)—conscious recollection *** Non declarative (implicit)—doesn’t require conscious recollection **** One type of implicit: procedural **** There are others (see large schematic) * Stages of memory: ** Encoding ** Storage/consolidation ** Retrieval * What are the sources of the evidence for these distinctions? Why do we use the categories that we do? ** Dissociations in learning and memory resulting from brain injury *** If brain region A is damaged, the patient maintains memory type X (e.g., long term) but not Y *** If brain region B is damaged, the patient maintains memory type Y but not X *** The fact that these memory types can be isolated (dissociated) in this way means that the brain carries them out separately, which is why we categorize them in this way. ** Dissociations in normal behavior resulting from brain injury (similar logic as above) Time-scale: Sensory, Short term, Long term * '''Sensory Memory (lasts the shortest amount of time) ** Large amounts of info available for brief periods, overwritten by incoming info ** Information available at low levels of sensory systems (e.g. V1; never goes up to V4) ** Types: *** Visual: iconic, 500 ms *** Auditory: echoic, several seconds ** Full/Partial report: much more info is available than revealed by full report, though it fades quickly while responding in full report. ** People generally only report 4 letters reliably. If you ask them to report the letters on the screen starting from the first line, they report the first 4 but can't remember the rest. Does this mean that only the first 4 letters reached your sensory memory? To find out, repeat the trial, but this time, ask the subject to report letters beginning from the second line of text. They again report 4 letters and cannot remember the rest, but this time it was beginning from the second line. This shows that all letters reached the sensory memory, not just the first 4. However, by the time you've finished reporting 4 letters (regardless of which line these letters are from), the rest have been discarded from your sensory memory. ' * Modal model: Atkinson & Shiffrin ** Hypothesis: info first held in STM and then transferred to LTM (STS gateway to LTS) ** Inputs --> sensory register --> short term storage --> long term storage ** Supported by neuropsych * KF: left''' hemisphere damage to perisylvian cortex (parietal-temporal lobes following motorbike accident) ** Lesion location for KF: after KF, many other reports of patients with left parietal/temporal lesions and short term memory deficits due to stroke, tumor ** Normal IQ/memory (even long term memory) except for digit span --> STM impairment ** '''Short term memory/Working memory *** Limited capacity store, used in maintaining for a brief period of time an active representation of info; seconds—minutes *** After info used, updated or discarded *** Examples of STM: Digit span task (read the first few sentences here if you are unfamiliar with the digit span task), to do lists, asking for directions 'o KF’s ability to learn&remember future and remember info from before implies that LTM is fine § KF STM was damaged! ' § This implies that LTM is independent of STM o Alternative hypothesis: STM and LTM independent learning and memory systems o KF disproves modal model''' - '''EP: amnesic; damage to Medial Temporal lobes; fine with STM o Bilateral medial temporal lobe damage''' o EP pretty good with direct copy of an image, but terrible with delayed recall o He had fine working memory and recall of old memories o However; '''impaired creating of new, lasting memories since the illness ' o Anterograde vs. retrograde amnesia § Anterograde: new, lasting memories § Retrograde: old memories o EP had anterograde amnesia' § New STM: intact''' § '''Old LTM: largely intact ' § New LTM: impaired' - New hypothesis: STM and LTM are relatively independent learning and memory systems Baddeley-Hitch model: working memory - Working memory': dynamic process where information is worked on' o Contrasts with passive nature of simple information store o 3 important characteristics differentiate the baddeley hitch model from modal model § Short term storage isn’t just way station to LTM § Central executive''' manipulates STM''' § 2 memory buffers: · Visuospatial scratchpad · Phonological loop o Phonological Loop <--> Central Executive <--> Visuospatial scratchpad o Phonological loop: acoustically coding info in WM § When subjects recall visual strings, they mix up “gee” and “tee”àimplies phonological coding, not visual coding o Visuospatial sketchpad: info storage in purely visual or visuo-spatial codes § E.g. remembering locations of flashing targets o Central executive: command/control center that coordinates interaction o When told to do arithmetic, you do a bunch of processes with WM § store and retrieve are memory processes''' § inhibit stim, applying rules, doing the actual operation, switching attention are executive processes o Remembering a string of numbers is a phonological loop task o Remembering colors and locations is a visuospatial sketchpad task o People are better at dual task of p.loop+v.sketch task than p.loop+p.loop task § More overlap in process, thus more interference § This suggests separate WM systems - Types of long term memory o '''HM: Bilateral medial temporal lobectomy (hippocampi) § Normal intelligence and perception and IQ § Normal digit span&STM § Severely impaired LTM: can’t form new, lasting memories (anterograde amnesia) § Temporally graded retrograde amnesia § Can’t remember a task or the experimenter § Can’t learn new words § Can learn and master new skillsà non-declarative o HM had deficit in creating declarative memories o Declarative is knowledge of facts and events; conscious knowledge § Episodic: episodes (events, time, place) § Semantic: world & knowledge o HM had difficulty acquiring both new episodic and semantic knowledge § Other amnesics had selective difficulties with episodic and semantic knowledge as well o HM was able to make new LTM for non-declarative procedural memory § He got better at tower of Hanoi even though he thought he was doing it for the first time each time ' o Procedural: knowledge of process(motor, perceptual, or cognitive) o HM had trouble with episodic and semantic declarative memories, but ok with procedural memories' - Implicit vs. explicit memory o Implicit (non-declarative): knowledge that can be retrieved without conscious recollection § Procedural is a type of implicit, along with: § Perceptual representation system (priming) § Classical conditioning (2 stim associated) § Nonassociative learning(habituation, sensitization) o Explicit (declarative): knowledge that can be retrieved only with conscious recollection o Neurologically intact individuals and amnesics show dissociations § Skill learning: improvement due to repeated practice · Intact individuals’ explicit memory do not report repeated sequence · However, reaction times are fasterà implicit memory working § Priming: IMPLICIT. '' · dissociation between explicit and implicit memory · Improvement in identifying/processing stimulus as a result of its having been processed previously · Facilitation due to repetition · Study a list of words and… o Explicit test: did you see this or not o Implicit test: win___ o Priming score = diff in likelihood of completing stem w/ window if you have studied vs have not studied the word · Priming effect: increased likelihood of completing stem w/ word from list · Recognition memory accuracy: explicit/conscious recollection of studied words · Priming effect was present even in EP (amnesic), but explicit/recognition memory went down for EP - Stages in memory o Encoding: acquisition of memory: experience change brain laying down memory trcaes o Storage (consolidation): retain memory traces over time o Retrieval: accessing stored memory traces - Encoding o Experience: went with ___ to ___ o Many features § Sensory § Conceptual –thoughts § Emotional o Sucessful encoding recquires: § Feature binding: features from single memory linked to feature pattern to ensure complete memory § Pattern separation: feature patterns from diff memories distinguished to avoid confusion of similar memories - Retrieval o 1. Retrieval cue: movie Y o 2. Pattern completion § Retrieval of features associated with movie Y (friend, theater) o 3. Pattern evaluation: veridical/fantasy, complete/incomplete § Criteria for evaluation · Degree of familiarity, vividness, detail False memories (confabulation) can be error in encoding or retrieval